Initializing help system before first use

Maximise discount at a bookstore


Type: discrete NLP
Rating: 2 (easy-medium)
Description: A bookstore has the following discount policy: For each USD 1 you spend you get 0,1 % discount on your next purchase.
Example: If you have to buy three books that cost USD 10 USD 20 and USD 30 you could buy the USD 30 book today, the USD 10 book tomorrow (on which you'll get a 3 % discount), and the USD 20 book the following day (on which you'll get a 1 % discount). Or you could buy the USD 30 book and the USD 20 book today, and the USD 10 book tomorrow (with a 5 % discount).
What is the cheapest way to buy N books (for given prices) ?
File(s): bookdisc.mos


bookdisc.mos
(!*********************************************************************
   Mosel NL examples
   =================
   file bookdisc.mos
   `````````````````
   A bookstore has the following discount policy: 
   For each USD 1 you spend you get 0,1 % discount on your next purchase.
   Example: 
   If you have to buy three books that cost USD 10 USD 20 and USD 30 you could 
   buy the USD 30 book today, the USD 10 book tomorrow (on which you'll get  
   a 3 % discount), and the USD 20 book the following day (on which you'll get 
   a 1 % discount). Or you could buy the USD 30 book and the USD 20 book today, 
   and the USD 10 book tomorrow (with a 5 % discount).
   What is the cheapest way to buy N books (for given prices) ?

   Based on AMPL model bookdisc.mos by M.J.Chlond
   Reference: J. & L. Poniachik, Hard-to-Solve Brainteasers (p16), Sterling

   (c) 2008 Fair Issac Corporation
       author: S. Heipcke, Jul. 2003, rev. Mar 2013
*********************************************************************!)

model "bookdisc"
 uses "mmxnlp"

 parameters
   N = 8                        ! Number of books
   D = 4                        ! Number of days
   DISC = 0,001                 ! Discount factor
 end-parameters

 declarations
  BOOKS = 1..N                  ! Set of books
  DAYS = 1..D                   ! Set of days
  COST: array(BOOKS) of real	! Cost of books
  ifbuy: array(BOOKS,DAYS) of mpvar  ! ifbuy[i,j]=1 if book i bought on day j, 0 otherwise
  total: array(DAYS) of mpvar	! Total cost of books on day j
  totcost: mpvar                ! Total cost of all books
 end-declarations

 forall(i in BOOKS,j in DAYS) ifbuy(i,j) is_binary
! forall(j in DAYS) total(j) is_integer

! Generate random book prices 
 setrandseed(3)
 forall(i in BOOKS) COST(i):= 10+round(50*random)     

! Objective: total cost
 totcost = sum(i in BOOKS) COST(i) -
           sum(k in DAYS | k>1) DISC*total(k-1)*total(k)
 
! Total cost per day
 forall(j in DAYS) total(j) = sum(i in BOOKS) COST(i)*ifbuy(i,j)
! Every book is bought on one day
 forall(i in BOOKS) sum(j in DAYS) ifbuy(i,j) = 1

! Solve the problem 
 setparam("xnlp_verbose",true)
 setparam("xnlp_solver",0)

 minimize(totcost)

! Solution printing
 TOTCOST:= sum(i in BOOKS) COST(i)
 writeln("Total cost: ", strfmt(getsol(totcost),5,2), " (orig: ", TOTCOST, ") discount:",
         strfmt((TOTCOST-totcost.sol)/TOTCOST * 100,5,2), "%" )
 forall(i in BOOKS) BDay(i):=round(getsol(sum(j in DAYS) j*ifbuy(i,j)))
 forall(i in BOOKS)
   writeln("Book ", i, " bought on day ", BDay(i), " at ",
           strfmt(COST(i)*(1-if(BDay(i)>1, DISC*total(BDay(i)-1).sol,0)),5,2),
           " (orig: ", strfmt(COST(i),5,2), ")" )
 
end-model